Method for developing a nonlinear, stationary or dynamical model of a control variable of a construction machine

摘要

The method of creating the machine model with respect to all operating points of the machine preferably uses neural networks. The control parameter is dependent on a set of input parameters esp. system parameters, system parameters differentiated over time, time delayed fed-back output parameters and/or predetermined values at the desired operating point. The control parameters for a sub-set of operating points are determined using measurement technology. Using a simplified partial model, one output parameter per model function is determined. At each desired operating point, the output parameters of each partial model function are weighted with a suitable weighting function and are combined to a total output parameter for that point. For all points, the difference between the total output parameter and the control parameter determined by measuring is determined. For operating points where this difference exceeds a predetermined value, a different model function or a different weighting function is used to bring the difference to a value below the predetermined value. The step of determining the difference and the step of using a different model function or weighting function are carried out as often as required for the statistically determined prediction quality of the entire model to reach the desired value.